afrexai-seo-content-engine
Warn
Audited by Gen Agent Trust Hub on Feb 14, 2026
Risk Level: MEDIUMPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION] (MEDIUM): High susceptibility to Indirect Prompt Injection. The skill's primary workflow involves 'real web search' and article scoring, which requires the agent to ingest untrusted content from the internet or user-provided documents. Attackers can leverage these ingestion points to embed instructions that override the agent's behavior during content generation or research tasks.\n
- Ingestion points: Web search results for keyword research and external articles for SEO scoring.\n
- Boundary markers: None identified; the provided documentation lacks delimiters or instructions to ignore embedded content.\n
- Capability inventory: Writing articles, planning content clusters, and generating schema markup.\n
- Sanitization: No evidence of sanitization or instruction-filtering logic was found.\n- [NO_CODE] (INFO): The skill identifies as 'No code' and contains no executable scripts or binaries. This eliminates the risk of traditional remote code execution, unauthorized network exfiltration, or persistence through malicious software, concentrating the security profile on the logic of the prompt instructions themselves.
Audit Metadata